Microsoft commits to user privacy in IE10; risks advertiser wrath over Do Not Track support

Microsoft has affirmed its commitment to shipping IE10 and Windows 8 with “Do Not Track” (DNT) enabled by default. In doing so, it’s set the stage for a major war over user privacy, and appointed itself an unofficial white knight of user privacy.

Feel free to pause a moment and let the cognitive dissonance fade.

For those of you just tuning in, Do Not Track is a proposed privacy flag whose implementation is currently being drafted by the W3C. As currently implemented, a browser with DNT activated notifies the target website that the user does not wish to be tracked. A website that supports DNT would (theoretically) honor the flag and agree not to track the visitor’s activities. DNT is a voluntary standard — there’s no federal or state law mandating its use and very few websites currently support it. If a website doesn’t support DNT, turning it on in the browser does nothing.

A number of advertisers have signed on to help draft the DNT standard in a bid to avoid a government-mandated solution that might be far more limiting, but that doesn’t mean they like it. Microsoft’s initial statement that DNT would be enabled by default came as a total surprise to many of the participants in the W3C’s draft process and was sharply criticized from certain quarters. In June, the W3C changed the DNT draft to propose that browsersmust ship DNT off by default and enable it only if the user gives “explicit consent.” This implied that if Microsoft went ahead and shipped IE10 with DNT activated, it wouldn’t be able to claim its browser supported DNT.

Microsoft’s unofficial response? “Bring it.” In a recent blog post, Brendon Lynch, MS’s Chief Privacy Officer, spelled out how DNT is implemented in IE10. Users who choose “Express Settings” for browser configuration at first-run will have their DNT flag enabled. Users who choose to customize their options will be prompted to set the flag manually. In both cases, users are told what the DNT flag is and how to change it if they choose to do so.

“Our approach,” Lynch writes, “is part of our commitment to privacy by design and putting people first. We believe consumers should have more control over how data about their online behavior is tracked, shared, and used.”

Ulterior motive?

A number of pundits have questioned whether or not Microsoft is using privacy issues to strike at Google’s advertising empire. The truth is, we don’t know. Some draft standards are widely adopted before the final version is complete — 802.11n was a good example of this — but with DNT, the implementation is still being worked on. Browsers that support DNT tend to do it in different ways, and websites are waiting for a better understanding of what they need to do before they go ahead with implementing it.

Microsoft’s decision to ship DNT as active by default isn’t actually going to mean much right away. It’s already led the W3C to consider modifying the DNT proposal; the business interests working on the standard are petrified that an opt-in network will destroy their companies and bombard users with constant requests for data sharing. Some of these concerns are shared by privacy advocates; a standard that destroys companies and creates an impenetrable tangle of pop-ups is in no one’s best interests.

Visit Wired with IE10 and DNT enabled, and this is what you’ll see

For the moment, I’m willing to take Microsoft at its word. What’ll be critical to watch is how the company implements DNT on its own websites, what it advocates as the standard evolves, and whether its positions take principled stands on defending users’ rights, or are drawn in a way that benefits itself while choking out the competition. This is a move that really could go either way. If Windows 8 catches fire, it’ll fundamentally change the way Microsoft interacts with its customers and users. It’s fair to give the company a chance to demonstrate the principles it intends to uphold as part of that change — but it’s also fair to keep a weather eye on what’s going on behind the scenes.

[Image credit]

Source

Is ‘cloud security’ an oxymoron?

Like it or not, technology is becoming ever more reliant on the cloud, and that has both positive and negative ramifications. On the positive side, cloud computing has opened up a whole new world of productivity that didn’t fully exist before services like Google Docs and Office 365 came into view. On a more personal scale, the cloud flipped the entertainment industry on its head with online music lockers that make it possible to access playlists and rock out to your ZZ Top albums wherever there’s an internet connection. Whether for work or play, living in the cloud can be truly awesome, but what happens when the sky starts falling?

Unfortunately, playing in the cloud isn’t always sunshine and rainbows, a point recently underscored when Mat Honan over at Wired fell victim to a band of hackers who weaseled their way into his Apple’s iCloud account. Honan recounted in frightening and fascinating detail how the cyber scoundrels took advantage of iCloud’s lack of a two-factor authentication process to access his account and remotely delete the data on his MacBook, iPad, and iPhone. Childhood pictures of his daughter and everything else he had stored on his mobile devices were wiped clean.

If you’re wondering how exactly this was made possible, Honan’s investigation, which involved chatting with one of the hackers involved, takes readers through a thrilling tale of how various online services — in this case, Amazon, Gmail, iCloud, and Twitter — reveal just enough personal information to pose a threat to one another if a hacker is truly determined. The devil isn’t in the details, however, it’s in our reliance on cloud computing, and sometimes there’s hell to pay.

What if I’m extra careful?

Millions of PlayStation Network subscribers saw the dark side to cloud computing when a massive security breach compromised their accounts.

Millions of PlayStation Network (PSN) subscribers trusted Sony with their name, address, birthday, credit card details, and other personal information, only to have all that information fall into the hands of hackers following a massive data breach that knocked PSN out of commission for almost a month. It was a rude awakening for millions and a reminder to us all that the concept of the cloud is still very much in its infancy and not without significant security risks.

Hacker attacks aren’t the only potential dark clouds. Computing in the cloud requires a level of a trust for things that are ultimately out of your control, like data outages and redundancy. If a data center goes up in flames, you have to trust that a team of engineers will be able to repair the damage quickly and efficiently, and that offsite backups exist so that your data can be restored. It happensmore often than many people realize.

What can I do?

Parking your data on third-party services isn’t inherently bad, it’s just risky. You can mitigate that risk by practicing smart computing habits, which first and foremost involves maintaining multiple backups. How diligent you should be depends entirely on how much you value your data, and if you’re sitting on mines of mission critical code, at minimum you should be backing up your digital bits to an offsite location on a weekly basis.

As for cloud computing services like Netflix, Dropbox, and anything else you might want to use, being safe is a two-prong process. First, do your research and find out what kind of security practices are in place before entrusting your personal information. As Honan found out, it’s comparatively easy to break into someone’s iCloud account and wreak havoc, and until Apple addresses this, you may want to avoid using it. There’s no shortage of cloud services available, so don’t settle on one that doesn’t take security as seriously as you do.

Secondly, it all comes full circle to safe computing practices, which we highlighted in a recent article. Unless you’re willing to unplug from the internet entirely (yeah, right!), you’re going to introduce some level of cloud-based risk into your online life. Safe computing practices ensure you’re doing everything possible on your end to prevent disaster, eliminating half the risk in the process. The other half is up to the players on the other side of the equation, and unfortunately, the very nature of cloud computing leaves that part out of your control.

Read: What would it feel like to live in the cloud?

Source

Don’t count your penguins before they’ve hatched

Leave it to Richard Stallman to pour cold water on a hot idea. The noted free software activist has come out against Steam for Linux, writing that “Nonfree game programs (like other nonfree programs) are unethical because they deny freedom to their users. (Game art is a different issue, because it isn’t software.) If you want freedom, one requisite for it is not having nonfree programs on your computer.”

It’s the argument I would have accepted from Stallman, actually. But it doesn’t mean it’s necessarily right — or the best one he could have made.

I’ve expressed here before that I vigorously support the rights of artists to get paid for their work, and I regard software the same way. My work in that realm has been relatively slight, but it’s been significant enough to convince me that the people who develop the ideas and execute the specific programming behind even the simplest applications deserve to make whatever they can from their efforts. What they do is hard, exacting work, which requires years of training, study, and practice, and giving them a little money in return — and the ability to specify (within certain restrictions) how that software may or may not be used or distributed does not seem that outlandish to me.

As Linux has shown time and time again over the last 20 years, with even the best free software, you typically get what you pay for: half-realized concepts, poor user interfaces, and support that’s either nonexistent or completely relegated to a community that may or may not be able to help you with whatever issue you’re facing. More elaborate software package (Windows, Office, most games) have tons of research, testing, and design muscle behind them. They’re so easy, and in many cases fun, to use because the companies have the money to ensure they are that way. That involves a lot of people and a substantial investment, and I have no problem with that being protected.

That said, I appreciate Stallman’s perspective, and how vocal he has been about it for a long time. He’s very much the type to practice what he preaches, and he’s been working to make software more open for decades. Even if I don’t agree with his every point, I absolutely think it’s a discussion worth having, and that the back and forth is, at its essence, generally good for the free software community. And, in an ideal world, Linux would be the completely free, completely open vehicle through which he could realize his dreams.

Of course, we don’t live in that world. In our world, the one in which Linux has less than 2% market share, such admirable idealism is not going to have the desired effect. It may even turn off many of the people who could potentially be interested in what free and open-source software has to offer.

The fact is that the people most likely to find Stallman’s argument powerful are those who are already committed to his cause. Everyone else needs to be convinced. And the only way that’s going to happen is if free software advocates entice them with something they already know they want and like — a type of digital gateway drug that will bring them into the fold. No, it probably won’t work as well as people like Stallman might hope, but it’s more likely to have a lasting impact than saying — through either word or implication — that people need to settle for inferior products. Which, if we’re being honest, is where Linux remains on most fronts, but especially with games.

Steam may encourage people to use paid software on Linux, but it will also force them to expand their horizons by experiencing firsthand what the operating system is capable of. Linux has a reputation of being limiting and difficult to use, and if it’s not entirely unearned, this could go a long way to helping the community prove how much the OS has evolved and improved in recent years.

If an ordinary guy used to using commercial software becomes frustrated by Windows 8, to pick a random (yet totally believable) example, and looks into one of the friendlier Linux distros (Ubuntu or Linux Mint, maybe?) as an alternative, you have to give him something or he won’t be willing to fully make the switch. Once that something is out there, then you can try to reel him in the rest of the way. “Sure, it’s great to be able to play all these games, but Linux is really about keeping software free and open. Why not try these titles that you don’t have to pay for, and that don’t saddle you with byzantine licensing restrictions?” Maybe he’ll accept it, maybe he won’t, but he’ll already be immersed in Linux, and will thus need a lot less persuasion than he might otherwise.

Get enough people on board — by whatever method — and software developers might see the value in Linux and be willing to devote more of their time and resources to creating software that will work with it. When that happens, everyone benefits. But the chances that it will ever happen when Linux commands as little public attention as it now does is, at its best, naïve. And if it has to be forever constrained in the ways Stallman suggests, the chances are excellent that it’s not going to happen at all.

To his credit, Stallman obviously understands this to some degree. He goes on to say in his piece that “if you’re going to use these games, you’re better off using them on GNU/Linux rather than on Microsoft Windows,” but it ultimately amounts to little more than a grudging endorsement of the lesser-of-two-evils variety. He continues: “If you want to promote freedom, please take care not to talk about the availability of these games on GNU/Linux as support for our cause.”

Freedom, as they say, isn’t free. There’s always a cost. For Linux that cost might be, at least temporarily, opening the doors to paid, closed software just to prove to people the possibilities of the platform. Stallman’s goal of ultimate freedom by way of Linux may be attainable, even within his lifetime. But it will never come to fruition as long as stalwarts like Stallman only emphasize their devotion to outfitting Linux with a different kind of chain.

Read: Could this be the year of the Linux desktop? and Valve: OpenGL outperforms DirectX — even on Windows

Source

Microsoft’s tweaks to Skype could facilitate wiretapping (Updated)

VoIP, the voice-over-IP communications technology that is slowly making POTS landlines obsolete. SIP providers, VoIP applications, and messaging platforms all utilize VoIP to provide voice calling on PCs, phones, and mobile devices. One of the most popular VoIP applications is the Skype messaging service. Skype uses a peer-to-peer network of internet nodes to route voice and/or video calls between users around the world. Especially in the case of consumer-grade VoIP, it is significantly cheaper than a traditional landline for voice calls, and it can potentially deliver better sound quality. Another area where VoIP services like Skype excel is as a communication medium for criminals. Thanks to the fast pace of technology and the use of a peer-to-peer connection, Skype is a decent platform to communicate without fear of others listening in, to an extent.

Of course, Skype is not a fully decentralized service because it uses so-called “supernodes.” The supernodes are basically servers that both the caller and recipient can connect to, and they use these mutually-known servers to make the initial introduction between the two clients. Reportedly, Microsoft is re-engineering these supernodes to make it easier for law enforcement to monitor calls by allowing the supernodes to not only make the introduction but to actually route the voice data of the calls as well. In this way, the actual voice data would pass through the monitored servers and the call is no longer secure. It is essentially a man-in-the-middle attack, and it is made all the easier because Microsoft -– who owns Skype and knows the keys used for the service’s encryption -– is helping.

As far as what this means for you, if you are not doing anything malicious then you don’t need to worry too much. Patriot Act exceptions aside, you would have to be acting suspiciously enough for a judge to grant a warrant before your conversations could be snooped. With that said, it is a bit disconcerting that it is possible to violate your privacy, especially when you aren’t doing anything to warrant such potential invasions.

Your best bet for securing your voice communications for the simple sake of privacy is to set up your own VoIP “softphone” with open source SIP software, and use end-to-end encryption and keys that you control access to. Such encryption includes ZRTP for the secure key exchange and SRTP for securing the voice (data) stream between you and the recipient. SRTP in particular is interesting because it uses, by default, a 128-bit key derived from a master key — exchanged using the ZRTP (or similar) protocol — that is further salted with a 112-bit key (which helps make the encryption key harder to brute force by making it more computationally expensive to do so).

The move by Microsoft is somewhat disheartening, but at the end of the day it will not affect the company’s userbase much. Yes, your conversations are potentially less private and secure, but Skype remains one of the easiest (and free) VoIP clients to use. Skype is now essentially equivalent to other traditional forms of communications like landlines and cellphones that are already capable of being tapped. From the perspective that it is a necessary evil to have to monitor and find malicious people, it is not a bad thing for Microsoft to do so long as it conforms to legal procedures and is not abused. That last part is, I think, what worries a lot of privacy conscious people, and if you do value security over convenience there are definitely better options out there than Skype.

Update – 3:33pm – Skype has contacted us to note that the changes were made in order to “improve the Skype user experience”, not to open the doors to tapping.

Regarding the supernodes Mark Gillett, Skype’s Corporate VP of Product Engineering & Operations noted:

As part of our ongoing commitment to continually improve the Skype user experience, we developed supernodes which can be located on dedicated servers within secure datacenters. This has not changed the underlying nature of Skype’s peer-to-peer (P2P) architecture, in which supernodes simply allow users to find one another (calls do not pass through supernodes). We believe this approach has immediate performance, scalability and availability benefits for the hundreds of millions of users that make up the Skype community.

And in response to that claim that the source code was leaked, Skype’s Chief Security Office, Adrian Asher, wrote:

Skype takes all necessary steps to prevent/defeat nefarious attempts to subvert the Skype experience. Skype takes its users’ safety and security seriously and we work tirelessly to ensure each individual has the best possible experience.

Of course a government wiretap is not something a corporation (or most people) would consider to be “nefarious”, but Skype has said to us that the changes were not made to help law enforcement.

Source

Threading the nanoscale needle: Applied Materials launches new interconnect technology

I’ve written quite a bit about the future of semiconductor design over the past eight months, but virtually all of our coverage has focused on materials engineering and the search for molecular structures that scale more effectively than conventional CMOS as we plunge below the 20nm node. Interconnects — the tiny wires that connect the various silicon layers inside a processor — are something we haven’t spent much time on to date. Today, Applied Materials is launching its new “Amber” PVD (Physical Vapor Deposition) technology for building interconnects.

One of the reasons that the ITRS — International Technology Roadmap for Semiconductors — is so ominous and foreboding is that the semiconductor industry faces enormous challenges in virtually every field. Interconnect technology is no exception; one of the problems Applied Materials is trying to solve is that conventional approaches to interconnect fill no longer work well as the gaps between contacts are too small.

As the aperture narrows and deepens, conventional plating leads to gaps — voids — in the fill layer. Such voids are difficult to directly detect, and circuits created with such flaws may initially function as designed. Because the amount of conducting material is significantly lower than it’s supposed to be, however, the copper at the edges heats more quickly, is subjected to greater electrical strain, and eventually fails. This is an obvious problem for device manufacturers, given that modern CMOS chips use literally miles of interconnects.

The graph in the upper-left-hand corner shows the length of the vias in a processor, the total number, and the number of single vias (vias that link two specific points). The lower-right-hand graph shows the impact of defective vias on semiconductor yields. At 130nm, high defect levels had a relatively small impact on overall yields. At 28nm, defect levels that wouldn’t have sabotaged 5% of yield at 130nm are crippling, killing upwards of 30% of a product. This has been partly driven by the explosive growth in the number of vias per product.

Applied Materials’ new method for interconnect fill is to deposit the material cold and then introduce heat. This reflow process leverages capillary action, which is defined as “the ability of a liquid to flow in narrow spaces without the assistance of, and in opposition to external forces like gravity… If the diameter of the tube is sufficiently small, then the combination of surface tension (which is caused by cohesion within the liquid) and adhesive forces between the liquid and container act to lift the liquid.”

Capillary action is the reason a paintbrush absorbs paint, it’s why paper towels work, and in this case, it’s how Applied Materials can build void-less interconnects at the necessary sizes without radically rethinking the manufacturing process. It’s not the perfectly-scaling, indefinitely applicable solution everyone would like to find — the general lack of such solutions is something of a theme in our coverage of semiconductor manufacturing — but it pushes back the boundaries while nanowires, air gaps, carbon nanotubes, and optical interconnects (both inter- and intra-chip) are all being researched. AppMat suggests that this approach will scale well into the 1Xnm process nodes, by which time problems in other areas may have made the interconnect issue trivial regardless.

The ITRS’s long-term outlook is rather cheerless. “It is clear that while transistor performance intrinsically improves with geometric scaling, interconnect performance does not. This implies that unless revolutionary interconnect solutions are found, interconnects will increasingly limit the performance and power efficiency of new products. Looking forward, a coherent vision for both global and local interconnects faces numerous challenges and few potential solutions.”

For now, the advantage to the new “Amber” PVD system is that it pushes back the ITRS timeframe for “No Known Solution” to various interconnect problems back to 2019 as opposed to 2016. Hopefully the additional three years of research will be enough time to find different solutions or alternate materials.

Read more about Applied Materials, and the quest beyond 22nm

Source

Cisco responds to unhappy users, reboots Connect Cloud, restores router functionality

Last week, Cisco rolled out its Connect Cloud service as an automatic update to routers with that particular function enabled — and triggered widespread user outrage in the process. The company’s Privacy Supplemental that accompanied the service implied that it would monitor users’ internet history, the Terms and Conditions stated that Cisco could kill user accounts for accessing certain content, and the only way to log into the router without a Connect Cloud account was to take it offline.

This collectively pissed off just about everyone, including yours truly. We’ve been in contact with Cisco since our initial story; the company has now released a follow-up blog post detailing the upcoming changes to the service and apologizing for how the rollout was initially handled. Here are some of the highlights:

• No more mandatory sign-ups — Connect Cloud will be an optional feature Linksys owners can choose to register for. Routers will default to standard (read: local) management. You will not need a Connect Cloud account to sign into or manage router functions.
• No automatic updates if the auto-update feature is turned off – Originally, the T&C contained a statement that Cisco could update routers where auto-update was deactivated, if the company deemed it necessary. Cisco has now stated it will not do this, and will revise its documentation accordingly.
• No monitoring or disconnects — I’ll quote these bits directly, since the terminology is important. “Cisco Connect Cloud and Cisco Linksys routers do not monitor or store information about how our customers are using the Internet and we do not arbitrarily disconnect customers from the Internet. The Cisco Connect Cloud service has never monitored customers’ Internet usage, nor was it designed to do so, and we will clarify this in an update to the terms of service. Cisco’s Linksys routers do not track or store any personal information regarding customers’ use of the Internet.”
• Router updates will continue to be offered for both CC-enabled devices and standard interfaces – Cisco will continue to offer firmware updates for both interfaces and types of users. Choosing the standard interface doesn’t mean you’ll be cut off from future security.

Cautiously optimistic

After having several conversations with Cisco representatives, I’m hopeful that the company is serious about these changes. Is it possible that this was a cynical land-grab to see what customers would or wouldn’t tolerate? Absolutely yes — and that’s why we’ll keep an eye on the situation and the documents in question. At the same time, we need to acknowledge that the company’s statements track nearly perfectly with a list of recommendations we made earlier this week.

Given the company’s quick response and rollback, it’s at least plausible that this was a case of over-enthusiasm combined with a poor understanding of what Linksys users most value about the product. In comments around the web, the phrase “it just works” came up repeatedly when users described why they were Linksys customers. My own preference for the brand is based on such statements; Linksys routers are broadly compatible workhorses that handle home routing tasks simply and require minimal maintenance. With Connect Cloud, Cisco is clearly hoping to boost its visibility, but the company’s first attempt to do so went too far.

Hopefully a more measured approach will achieve the desired goal without the negative feedback. As an optional service, there’s nothing wrong with Cisco’s Connect Cloud. We understand that readers will be dubious after last week’s events; we’ll keep an eye on Cisco’s documentation and T&C and notify you of any changes.

Source

Facebook invests in 55-terabit intra-Asia submarine cable system

Facebook has joined a consortium that will build by far the fastest intra-Asia submarine fiber optic network, the Asia Pacific Gateway (APG). Facebook is the only American company involved with the venture, which will see 10,000km (6,000 miles) of prime fiber laid between Malaysia and Japan (pictured above), with branches landing in almost every country along the way (Singapore, Thailand, Vietnam, China, Taiwan, and South Korea).

When the cable goes online in 2014, it is slated to use 40Gbps channels, for a total capacity of 55 terabits per second, or a transfer speed of 6.9 terabytes (138 Blu-ray discs) per second. When the various routers and repeaters are upgraded to 100Gbps-per-channel, the cable will have a total capacity of well over 100Tbps. The members of the consortium have put forward a total of $450 million so far, which makes it one of the most expensive submarine cable systems in the world.

The prime reason for building the cable, other than making money, is to provide more redundancy between the US and Asia. Currently, almost every connection from Asia to the US is routed through Singapore or Japan. If there is congestion at one of these sites, or a cable is cut, then the other (highly populous) countries in south east Asia are in trouble. It is perhaps no surprise that China Telecom and China Unicom, two of the world’s largest telecoms companies, are involved in the laying of APG. Time dotCom, another member of the APG consortium, says that the new cable will reduce latency between the US and Asia — which is fairly important, considering how many internet services are based in the US.

Beyond redundancy, this cable (and Facebook’s involvement) is significant for one very important reason: Over the next few years, almost all of the internet’s growth will occur in Asia. The total number of internet users has almost stagnated in Europe and North America — while in China, India, and south east Asia, hundreds of millions of people are connecting to the internet every year. Facebook has almost reached saturation point in Europe and North America, while its growth in Asia is only just starting to kick in. Backing this cable is Facebook’s way of acknowledging that Asian users are important to the continued growth of the social network — and, of course, the company is also communicating to its shareholders that its flagging US and EU growth isn’t a cause for concern.

While it seems quite sensible, it’s actually quite uncommon for web-based companies to invest in infrastructure beyond their own data centers. Google is one of the few exceptions, owning large swathes of the internet’s backbone, and being a member of the consortium that laid the Japan-US Unity submarine cable system. By owning a large portion of the network between its servers and its users, and thus control of the network setup and topology, it can provide much faster (lower-latency) access to its services. Being the most popular destination on the web, I wouldn’t be surprised if this marks the beginning of significant infrastructure investment by Facebook.

Read more about the secret world of submarine cables, or about the $1.5 billion polar Arctic Fibre and Arctic Link cables that will cut London-Tokyo latency by 60ms.

Source

US retakes supercomputing crown with 16-petaflops Sequoia; China promises 100 petaflops by 2015

Almost 2,500 high-performance computing boffins have descended upon the German city of Hamburg to attend the International Supercomputing Conference 2012 (ISC) this week to discuss the latest and greatest innovations and applications in the realm of supercomputing. The biggest news of ISC, by far, is that the USA has beaten back the Chinese and Japanese to reclaim pole position on the 39th Top500 list — the list of the world’s fastest supercomputers. In retaliation, China is now promising to deliver a 100-petaflops by 2015 — some two years before the rest of the world is expected to reach such lofty computational heights.

The world’s fastest supercomputer is now Sequoia, an IBM Blue Gene/Q installation at the Lawrence Livermore National Laboratory in California, primarily for use by the National Nuclear Security Administration. The system’s Linpack benchmark performance is 16.32 petaflops (16 quadrillion floating point operations per second), some 55% faster than the Japanese K, the previous record holder. The Sequoia’s theoretical max performance is just over 20 petaflops. Sequoia will be used to simulate the safety and efficacy of the US nuclear weapons stockpile, without the need for actual underground testing — and other scientific research.

Under the hood, Sequoia boasts 98,304 18-core processors and 1.6 petabytes of RAM, spread out amongst 96 racks that occupy a total floorspace of 3,000 square feet (280 square meters). Compute Node Linux powers around 97,500 of the processors, and Red Hat Enterprise Linux runs on 768 I/O (filesystem) nodes. The Blue Gene/Q chip itself is a terrifying 1.6GHz 64-bit PowerPC A2, which is 4-way simultaneously multithreaded, with each core harboring a SIMD quad-vector double precision floating point unit. There are 18 cores in total, but one core is a backup in case another is permanently damaged, and another handles interrupts, asynchronous I/O, and other operating system helper functions. The chip is theoretically capable of 205 gigaflops while drawing 55 watts — just a wee bit more efficient than desktop PC chips.

The chip’s frugal power consumption means that Sequoia uses only 7.9 megawatts of power while running Linpack, or around two gigaflops per watt. This is enough to put Sequoia at the top of the Green500 list of most efficient supercomputers. In comparison,the K supercomputer uses 12.6 megawatts to perform 10.5 petaflops — or only 800 megaflops per watt. An Ivy Bridge Core i7-3770K achieves around 50 gigaflops while consuming around 100 watts — or 500 megaflops per watt.

Ni hao

Back in 2010, the 2.5-petaflops Tianhe-1A became the first Chinese installation to reach the top of the Top500, knocking the 1.7-petaflops Cray XT5 Jaguar supercomputer at the US Oak Ridge National Laboratory down into second place. This in itself was enough to get US alarm bells ringing — the US has basically been the undisputed king of supercomputing, except for a short period in the ’90s — and then Japan turned on the K computer at the RIKEN institute. K was four times faster than Tianhe and 6 times faster than Jaguar. The US has been in second place a few times throughout the years, but third place? Unthinkable!

Judging by the rumblings being made by the director of China’s National Supercomputing Center, though, perhaps America should get used to this being a three horse race. Speaking to The Register, Liu Guangming says that China will power up a 100-petaflops supercomputer sometime in 2015. This is rather impressive, as previous estimates have centered around the 2017-2018 mark.

Guangming says his team is currently evaluating what architecture the supercomputer will use. It could use Intel Xeons (like the Tianhe-1A), AMD Opterons, or its own homegrown CPUs. As we’ve reported previously, China has already built a 1-petaflops computer from its domestic 16-core 64-bit ShenWei CPUs. The Chinese also have to decide on whether the 100-petaflops computer will be a CPU-GPU hybrid (like Tianhe, which uses 7,168 Nvidia Tesla GPUs) or a straight-up CPU-powered beast. Again, Guangming says that they might use homegrown GPUs — though he didn’t share any details on their specs.

Back at home, the next big American supercomputer will be the Cray XK6 Titan, Jaguar’s successor. Titan, built out of AMD Opterons and Nvidia Kepler GPUs, should have a peak performance of 24 petaflops. You can see on Top500 that the upgrade process has already begun!

In other news, IBM’s SuperMUC, installed in Munich, Germany took fourth place on the Top500 and became the fastest supercomputer in Europe. SuperMUC is novel because it uses a hot water cooling system to reduce its power consumption by 40%.

Read about the history of supercomputers, or what supercomputers are actually used for

Source